Changing Mortality Trends in Countries and Cities of the UK: a Population-Based Trend Analysis
Total Page:16
File Type:pdf, Size:1020Kb
Open access Original research BMJ Open: first published as 10.1136/bmjopen-2020-038135 on 5 November 2020. Downloaded from Changing mortality trends in countries and cities of the UK: a population-based trend analysis David Walsh ,1 Gerry McCartney ,2 Jon Minton,2 Jane Parkinson ,2 Deborah Shipton,2 Bruce Whyte1 To cite: Walsh D, McCartney G, ABSTRACT Strengths and limitations of this study Minton J, et al. Changing Objectives Previously improving life expectancy and all- mortality trends in countries and cause mortality in the UK has stalled since the early 2010s. ► We examine recent changes in mortality in the UK in cities of the UK: a population- National analyses have demonstrated changes in mortality based trend analysis. BMJ Open the context of much longer term trends: almost 40 rates for most age groups and causes of death, and with 2020;10:e038135. doi:10.1136/ years in most cases. deprived populations most affected. The aims here were to bmjopen-2020-038135 ► Given the importance of urban health to national establish whether similar changes have occurred across outcomes, we include the largest cities in Scotland, ► Prepublication history and different parts of the UK (countries, cities), and to examine England (London excepted) and Northern Ireland. additional material for this paper cause- specific trends in more detail. ► We analyse data for 10 major causes of death (not is available online. To view these Design Population- based trend analysis. files, please visit the journal just all causes combined), and by country-specific Participants/setting Whole populations of countries and online (http:// dx. doi. org/ 10. and (for Scotland) city- specific deprivation levels. selected cities of the UK. 1136/ bmjopen- 2020- 038135). ► Limitations include the fact that interpretation of Primary and secondary outcome measures European trends at city level can be problematic, given the age- standardised mortality rates (calculated by cause of Received 28 February 2020 fluctuation in rates. Revised 10 October 2020 death, country, city, year (1981–2017), age group, sex Accepted 16 October 2020 and—for all countries and Scottish cities—deprivation quintiles); changes in rates between 5- year periods; summary measures of both relative (relative index changing trends have been observed in many of inequality) and absolute (slope index of inequality) other high- income countries, although the inequalities. slowdown has been particularly marked in Results Changes in mortality from around 2011/2013 5 the UK and the USA. Other high-income http://bmjopen.bmj.com/ were observed throughout the UK for all adult age countries with higher life expectancy have groups. For example, all- age female rates decreased 2 5 by approximately 4%–6% during the 1980s and 1990s, seen continued improvements. approximately 7%–9% during the 2000s, but by <1% Data from Scotland and England (92% of between 2011/2013 and 2015/2017. Equivalent figures the UK population) have shown that these for men were 4%–7%, 8%–12% and 1%–3%, respectively. changing mortality patterns have been This later period saw increased mortality among the most observed for almost all age groups and for deprived populations, something observed in all countries most causes of death.6 7 Worryingly, increasing and cities analysed, and for most causes of death: mortality rates among the most socioeco- on November 13, 2020 by guest. Protected copyright. absolute and relative inequalities therefore increased. nomically deprived populations have also Although similar trends were seen across all parts of the been observed; as a result, inequalities in all- UK, particular issues apply in Scotland, for example, higher cause mortality have widened considerably and increasing drug- related mortality (with the highest since around 2012.7–9 An emerging body of rates observed in Dundee and Glasgow). 1 10–13 14 15 Conclusions The study presents further evidence of UK and international work suggests © Author(s) (or their changing mortality in the UK. The timing, geography and the recent stalling is likely to be associated employer(s)) 2020. Re- use socioeconomic gradients associated with the changes with the implementation from 2010 of UK permitted under CC BY- NC. No appear to support suggestions that they may result, at Government ‘austerity’ measures—cuts to commercial re- use. See rights least in part, from UK Government ‘austerity’ measures and permissions. Published by public services and social security—which BMJ. which have disproportionately affected the poorest. have particularly affected the most vulnerable 1Glasgow Centre for Population populations. Health, Glasgow, Scotland, UK The principal aim of this project was to 2Public Health Scotland, INTRODUCTION establish whether similar changing mortality Glasgow, Scotland, UK The recent (pre-COVID-19 pandemic) slow- trends (in terms of rates of improvement, Correspondence to down in improvement in life expectancy and causes of death and socioeconomic inequal- Dr David Walsh; mortality rates in the UK has been highlighted ities) have occurred ubiquitously across the david. walsh. 2@ glasgow. ac. uk by researchers1 2 and media3 4 alike. Similar UK. This included examining cause-specific Walsh D, et al. BMJ Open 2020;10:e038135. doi:10.1136/bmjopen-2020-038135 1 Open access BMJ Open: first published as 10.1136/bmjopen-2020-038135 on 5 November 2020. Downloaded from trends in more detail, and focusing on selected individual Measure (NIMDM),20 respectively. In all three cases data cities, given the importance of urban health to national were available for the period 2001–2017. The SIMD has outcomes.16 been updated multiple times: thus, the 2004 version was used for analyses covering the years 2001–2004, SIMD 2006 was used for 2005–2007, SIMD 2009 for 2008–2010, METHODS SIMD 2012 for 2011–2013 and SIMD 2016 for 2014–2017. Mortality and population data Similarly, the (English) IMD 2004 was used for the years Numbers of deaths by year of registration, age, sex, 2001–2005, IMD 2007 for 2006–2008, IMD 2010 for 2009– underlying cause, city and country for Scotland, England 2013, IMD 2015 for 2014–2016 and IMD 2019 for 2017. & Wales and Northern Ireland were obtained from, For Northern Ireland, NIMDM 2010 was used for all years respectively, the National Records of Scotland (NRS), the of analyses. Although there are differences in the spatial Office for National Statistics and the Northern Ireland scale and the individual variables used in the construc- Statistics and Research Agency. Data were available for tion of each nation’s deprivation measure, all three share the following years: 1974–2017 (Scotland); 1981–2017 notable similarities in terms of their basic composition. (England & Wales); and 1997–2017 (Northern Ireland). The principal ‘data domains’ of each are effectively the Data were obtained for all-cause deaths, and for same: income; employment; health; education, skills and the following 10 major individual causes: respiratory training; crime; access to services; housing. In the Scot- disease; ischaemic heart disease (IHD); cerebrovascular tish index, housing is a separate category; in the English disease; all malignant neoplasms; lung cancer (malignant and Northern Irish indices it is instead contained within neoplasm of trachea, bronchus and lung); intentional a ‘living environment’ domain. For all three measures self- harm (including events of undetermined intent); of deprivation, similar methodologies are employed to external causes; motor vehicle traffic accidents (MVTAs); calculate an overall index of relative deprivation, based alcohol- related causes; and drug- related poisonings. on geographical area rankings across all data domains. Causes were defined by groups of International Classifica- Although the absolute values of the different indices tion of Diseases 8th Revision (ICD-8), ICD-9 and ICD-10 cannot be directly compared, the similarity of compo- codes: these are listed in the online supplemental table 1. sition and methodology associated with each provides As stated in online supplemental table 1, the definition of helpful, and broadly comparable, overviews of inequality external causes overlaps with other causes of death, that within each setting. is, MVTAs, intentional self- harm, drug- related poison- ings. ICD-9 codes were used for the years 1979–1999 in Statistical analyses Scotland, and for 1981–2000 in England & Wales and European age- standardised mortality rates per 100 000 Northern Ireland; ICD-10 codes were used for all later population were calculated using the 2013 European 21 years. ICD-8 codes were used for 1974–1978 (Scotland Standard Population. Analyses were undertaken by sex, http://bmjopen.bmj.com/ only) but the data for those years are not presented here. age (all ages, 0–64 years (the latter to examine prema- Matching population data by year, 5- year age group, sex, ture deaths) and four broad groups across the life course: city and country were obtained from the same national 0–14, 15–44, 45–64 and 65+ years), year, cause of death, statistical agencies. city, country and deprivation quintile (see below). Three- year rolling average rates were derived; to quantify the Geography rate of improvement over time, the percentage changes Scotland, England & Wales (combined) and Northern in rates between 3- year averages at 5- year intervals (ie, Ireland were the countries used in the main analyses. between 1981/1983 and 1985/1987, 1986/1988 and on November 13, 2020 by guest. Protected copyright. For analysis by deprivation quintiles (discussed further 1990/1992… up to 2011/2013 and 2015/2017) were below), England alone, rather than England & Wales, was calculated. Three- year averages were used to overcome used. With the exception of London, the largest cities in the issue of fluctuating rates (especially at city level). each country were selected: Glasgow, Edinburgh, Dundee For simplicity, we use the expression ‘five year’ interval and Aberdeen in Scotland; Liverpool, Manchester, to reflect the midpoints of the 3- year average (eg, 1982– Birmingham, Leeds, Sheffield and Bristol in England; 1986 in relation to 1981/1983 to 1985/1987).